Role of the Extensin Superfamily in Primary Cell Wall Architecture

Nearly 2 centuries of progress have established the major components of the plant cell wall, a composite that includes interpenetrating networks of cellulose (Payen, 1838; Schulze, 1891), microfibrils (Frey-Wyssling et al., 1948; Preston et al., 1948), pectin (Braconnot, 1825), and lignin (Payen, 1838). However, only over the last 5 decades has a relatively minor Hyp-rich structural glycoprotein component emerged with essential roles in building and maintaining the growing primary cell wall. Here, we highlight unique advances of each decade, from the initial discovery of Hyp in cell walls to the current definition of extensins as self-assembling amphiphiles that generate scaffolding networks, where acid-base interaction (extensin pectate) may template assembly of the pectic matrix. Subsequent polymerization toughens up the wall as networks resisting both microbial and mechanical stress. At each stage, we explore hypotheses arising from the synthesis of emerging data with focus on structure. This review celebrates the 50th birthday of extensin. Protein interactions direct life processes at all levels, ranging from the regulation of metabolism and nucleic acid replication to signal transduction and morphogenesis. Sophisticated extracellular matrices like those of animals and plants constitute scaffolding networks of glycoproteins and proteoglycans that interpenetrate other networks of structural polysaccharides. While polysaccharide networks are prominent in the plant extracellular matrix, glycoproteins and proteoglycans dominate the animal matrix. By weight, protein, which is largely structural, contributes up to 20% of the primary wall (Burke et al., 1974; Kieliszewski et al., 1992a) yet may be essential, as the loss of the structural glycoprotein network in Arabidopsis (Arabidopsis thaliana) is lethal (Hall and Cannon, 2002; Cannon et al., 2008). Since the discovery of cell wall protein in 1960 (Dougall and Shimbayashi, 1960; Lamport and Northcote, 1960b), the hydroxy-Pro-rich glycoprotein field, referred to generically as HRGPs, has blossomed from having only one family member (extensin; Lamport, 1963), to three (extensin, arabinogalactan protein [AGP; Yariv et al., 1962; Aspinall et al., 1969; Fincher et al., 1974], and solanaceous lectins [Allen and Neuberger, 1973]), then four (Pro-rich proteins, [PRPs; Chen and Varner, 1985; Hong et al., 1987; Averyhart-Fullard et al., 1988; Tierney et al., 1988; Wilson et al., 1994]) and more. We now realize that this (super)family represents a continuum of peptide periodicity and glycosylation (Kieliszewski and Lamport, 1994); perusal of the Arabidopsis genome suggests that conserved features of HRGPs, in particular their arabinosyl-O-Hyp and arabinogalactosyl-O-Hyp glycomodules, are widespread in secreted proteins ranging from phytocyanins and systemins to fasciclins and glycerophosphodiesterases (Pearce et al., 1991; Borner et al., 2002; Johnson et al., 2003; Kieliszewski et al., 2010; Showalter et al., 2010). Furthermore, highly organized HRGP cell wall networks are ancient, as they occur in algae (Thompson and Preston, 1967; Gotelli and Cleland, 1968; Miller et al., 1972; Roberts, 1974; Goodenough and Heuser, 1989). Here, we largely confine ourselves to the classical extensins, which we define as wall-located, basic, Hyp-rich structural glycoproteins with alternating hydrophilic and hydrophobic motifs whose alignment as self-assembling amphiphiles likely drives extensin network assembly. The hydrophilic motifs comprise arabinosylated X-Hypn, where X is usually Ser and n is most often 4 and occasionally up to 5 (Campargue et al., 1998) or six (Qi et al., 1995), while the hydrophobic “motifs” vary, sometimes being as small as a single amino acid or dipeptide and often containing Tyr residues as potential cross-link sites. The findings and technologies used are presented in a historical context, building up stepwise a picture of what we know about extensins, with focus on the big question: what role does extensin and a protein network play in cell walls? In 1960, the primary cell wall isolated from sycamore (Acer pseudoplatanus) cell suspension cultures and tobacco (Nicotiana tabacum) callus (Lamport and Northcote, 1960a) contained enzymes and a Hyp-rich component (Lamport and Northcote, 1960b); therefore, it could be considered as a “cell particle (or organelle) possessing structural integrity and enzymic autonomy” (Lamport, 1964). Hyp indicated a structural protein by analogy with animal collagen, where this cyclic amino acid constrains side chain rotation and yields an extended structural protein. Sycamore cells grown in 18O2 showed that molecular oxygen was the direct source of the hydroxyl oxygen (Lamport, 1963), which was significant because the Hyp hydroxyl plays a pivotal role as a carbohydrate attachment site (Lamport, 1967) and Hyp-rich glycopeptides isolated from enzymatic digests of tomato (Solanum lycopersicum) cell walls indicated a highly glycosylated glycoprotein (Lamport, 1969). Prior to the discovery of the endoplasmic reticulum/Golgi role in protein secretion (Jamieson and Palade, 1967), early pulse-chase experiments suggested that extensin destined for the wall of sycamore cells occurred as a soluble cytoplasmic precursor (Lamport, 1965). Later pulse-chase experiments showed that macromolecular Hyp appeared in “membranous organelles” before secretion to the wall (Chrispeels, 1969). A hypothetical role in cell extension based on the structure and location of Hyp-rich glycoprotein in the primary cell wall suggested the name “extensin” (Lamport, 1963). Identification of Hyp in the walls of many algae (Gotelli and Cleland, 1968) supported the hypothesis that extensins are widespread and may play a role in cell expansion (Thompson and Preston, 1967). Facile cleavage of acid-labile (pH 1) arabinofuranoside linkages and subsequent tryptic degradation released significant amounts of Hyp-rich material from the wall (Lamport, 1974) but only a very few major peptides each containing the diagnostic Ser-Hyp4 signature sequence. This was the first suggestion that extensin is a highly periodic protein, subsequently corroborated by the circular dichroism spectra of both crude extensin and extensin peptides, indicating an extended left-handed poly-Pro-II helix (Lamport, 1977). The intractability of the presumed extensin network provided the impetus for developing new tools, particularly hydrogen fluoride-solvolysis deglycosylation of glycoproteins, in an attempt to solubilize wall-bound extensin (Mort and Lamport, 1977) from putative glycan cross-links (Keegstra et al., 1973). However, extensin remained insoluble, indicating protein-protein cross-linking rather than protein-glycan cross-linking and confirming earlier work (Lamport, 1965). As the structure of extensin was being determined, their involvement in disease and wound responses became apparent when Esquerre-Tugaye and colleagues showed that pathogens induced extensin accumulation and that this was correlated with disease resistance (Esquerre-Tugaye and Mazau, 1974; Esquerre-Tugaye and Lamport, 1979), while Chrispeels et al. (1974) showed that physical wounding induced extensin biosynthesis. A general role for extensin in response to different stresses, including senescence and abscission, was corroborated and detailed (Merkouropoulos and Shirsat, 2003). The quest for salt-extractable monomeric precursors to network extensin began in the early 1960s, but low yields from sycamore cell suspensions (Lamport, 1965) and carrot (Daucus carota) discs (Brysk and Chrispeels, 1972; Stuart and Varner, 1980) impeded progress. Finally, substrate quantities of extensin monomers salt eluted from intact cells of rapidly growing tomato cell suspension cultures (Smith et al., 1984) allowed detailed characterization that confirmed the remarkable periodicity of the Ser-Hyp4 glycomotif and also the precursor-product relationship between monomeric extensin and the insoluble wall network. Salt elution also implied ionic interaction between the extensin and pectin networks (Smith et al., 1984, 1986; Qi et al., 1995; Nuñez et al., 2009). Possession of a substantial monomeric pool enabled in vitro cross-linking experiments. The discovery of the cross-link amino acid isodityrosine (Idt) in cell wall hydrolysates (Fry, 1982) sparked speculation that Idt was the intermolecular cross-link and key to extensin network insolubilization (Fry, 1982; Lamport and Epstein, 1983). However, the insoluble extensin wall network yielded tryptic peptides that contained Idt only as a very short intramolecular cross-link in a highly conserved hydrophobic motif, Tyr-Xaa-Tyr (Epstein and Lamport, 1984). Nevertheless, the idea of Idt intermolecular cross-links persisted, fueled by further evidence of in muro cross-linking (Cooper et al., 1987; Bradley et al., 1992). In particular, Bradley et al. (1992) showed that fungal elicitation of hydrogen peroxide corresponded to a rapid wall-hardening process involving a decrease in extractable extensin, emphasizing the significance of Esquerre-Tugaye’s earlier work (Esquerre-Tugaye and Mazau, 1974; Esquerre-Tugaye and Lamport, 1979) and the highly specific pI 4.6 extensin peroxidase that catalyzed in vitro extensin cross-linking (Everdeen et al., 1988; Lamport, 1989). Evidence of other extensin peroxidases appeared later (Price et al., 2003). Finally, the diagnostic Ser-Hyp4 peptide (Smith et al., 1986) enabled identification of the first extensin (Chen and Varner, 1985) and PRP cDNAs (Hong et al., 1987; Tierney et al., 1988; Datta et al., 1989) as bona fide proteins with the hallmark of other structural proteins, most notably collagen, which is also Hyp rich and the major structural fibrillar protein of animals. Collagen polypeptides occur in an extended poly-Pro-II left-handed helical conformation, which was also confirmed in carrot extensin by further circular dichroism spectra (van Holst and Varner, 1984), with evidence for the role of carbohydrate in maintaining the backbone conformation (Stafstrom and Staehelin, 1986). Evolution conserves functional motifs. Peptide sequence motifs from gymnosperms (Fong et al., 1992; Kieliszewski et al., 1992a) and dicot extensins (Smith et al., 1986; Li et al., 1990; Memelink et al., 1993) made a comparison with other advanced angiosperm groups of great interest, particularly those with a radically different growth habit, like the grasses. A Thr-rich HRGP (THRGP) from maize (Zea mays; Kieliszewski and Lamport, 1987; Hood et al., 1988; Stiefel et al., 1988) was clearly related to dicot extensins and suggested that the HRGP conserved sequence encodes both Pro hydroxylation (Kieliszewski et al., 1990) and Hyp glycosylation. Another HRGP from maize contained both extensin and AGP peptide motifs and led to the formulation of the Hyp contiguity hypothesis: “perhaps sequences around noncontiguous Hyp direct Hyp-arabinogalactosylation, whereas contiguous Hyp directs arabinosylation” (Kieliszewski et al., 1992a, 1992b). Such codes implied that extensin could readily evolve into an AGP or vice versa, as a single base change relates Pro, Ser, and Ala codons, respectively, CCX → UCX → GCX. Thus, a single base change transforms contiguous to noncontiguous Hyp, changing the glycosylation code, and explains why members of the extensin superfamily appear as a phylogenetic continuum (Kieliszewski and Lamport, 1994); for example, gum arabic glycoprotein possesses both contiguous Hyp (extensin) and noncontiguous Hyp (AGP) motifs (Qi et al., 1991). Synthetic gene constructs confirmed the Hyp glycosylation code and enabled the design of HRGPs to elucidate posttranslational codes and the function of conserved motif repeats (Shpak et al., 1999) in the following decade. Meanwhile, the significance of putative cross-link motifs, VYK and Idt (YXY), became apparent with the availability of FPLC, notably Superose-6 columns that allowed the resolution of extensin monomers, oligomers, and polymers following in vitro enzymatic cross-linking of the monomers (Schnabelrauch et al., 1996), and led to the discovery of a specific extensin peroxidase. Significantly, extensin peroxidase did not cross-link extensins like the maize THRGP (Schnabelrauch et al., 1996), which lacked the putative VYK and YXYK cross-link motifs. Some extensins, like tomato P1, ap parently lacking Idt motifs (Smith et al., 1984, 1986), were readily cross-linked, suggesting VYK as an intermolecular cross-link. However, discovery of the cross-linked Tyr derivatives di-isodityrosine (di-Idt) and pulcherosine (Brady et al., 1996) resolves the issue (Fig. 1). The tetra-Tyr derivative, di-Idt, can be formed from two Idt residues in neighboring molecules, while the tri-Tyr derivative, pulcherosine, can be formed from Idt and Tyr (Fry, 1982; Brady et al., 1996; Brady and Fry, 1997). The apparent absence of Idt motifs in the P1 peptides isolated earlier (Smith et al., 1986) may reflect the restriction of Idt to the C-terminal YVYSSPPPPYHY (SGN-U315189). Thus, abundant “non-Idt” Tyr residues in P1 and the two Idt motifs at the C terminus are sufficient for peroxidatic cross-linking to give pulcherosine and some di-Idt. In addition to specific motifs and their abundance, differential localization in the wall may also influence the role of extensins (Swords and Staehelin, 1993). Amino acid structures of Tyr derivatives. A, The diphenylether Idt. B, The tri-Tyr pulcherosine. C, The tetra-Tyr di-Idt. Glycoproteins smothered in sugar offer technical and conceptual challenges. How can we relate structure to function? In particular, what is the role of O-Hyp glycosubstituents: highly conserved neutral oligoarabinoside glycomodules, typically tri- and tetra-arabinosylated Ser-Hyp4 in extensins (Lamport et al., 1973), and the acidic arabinogalactan polysaccharide glycomodules of AGPs (Tan et al., 2004)? Both are hydrophilic, but their different structures imply different roles (Tan et al., 2010). Synthetic gene technology and molecular genetics have yielded insights into the assembly of the extensin network at the molecular level and its role at the biological level. Discovery of the lethal rsh embryogenic Arabidopsis mutant corresponding to AtEXT3 showed that extensins are essential for cell plate formation, evidenced by the aberrant mutant wall phenotype and AtEXT3 immunocytochemical localization (Hall and Cannon, 2002). At the molecular level, purified AtEXT3 extensin monomers visualized by atomic force microscopy (AFM) This a for by the alternating hydrophilic and hydrophobic motifs, typically Idt (Epstein and Lamport, 1984), and is with their in vitro by extensin peroxidase et al., 2008). AtEXT3 yielded the tri-Tyr pulcherosine as the major intermolecular cross-linked rather than formed in the et al., This was by comparison of AtEXT3 with the of the motifs in the sequence because all Tyr residues to Idt they di-Idt. cross-linked of AtEXT3 can only when the (Idt) motifs are in However, when the AtEXT3 alignment is pulcherosine is the cross-linked as an Idt motif only with single Tyr residues or in neighboring AtEXT3 et al., 2008). Such AtEXT3 alignment is with the in vitro cross-linking and with the of AtEXT3 as by is a general of extensins, by of a of extensins (Fig. example, tomato extensin P1 and the maize THRGP network THRGP is not cross-linked by extensin peroxidase (Schnabelrauch et al., 1996), we that hydrophilic alternating with hydrophobic motifs monomers into of to networks where and C-terminal sequences and network assembly and may also (van 2010). Thus, hydrophilic arabinosylated and motifs of THRGP with hydrophobic Tyr residues and the hydrophobic motifs and et al., of extensin networks that rapidly but not extensins et al., are also in cell and the of new of self-assembling tomato maize were in to a of of the protein was a highly for 5 and then The was with of and then was on an in The P1 is a while the THRGP is a further and comparison with of an Arabidopsis extensin, Cannon et al. The of self-assembling amphiphiles very likely to interactions in the wall. interaction between extensin and pectin (Smith et al., 1984) has the potential to extensin et al., 2008). Thus, extensin may template the assembly of pectin in the cell involving some cross-links (Qi et al., 1995; Nuñez et al., 2009). AGPs may also as self-assembling amphiphiles that are into the in an et al., where they are by their C-terminal hydrophobic and et al., Borner et al., 2002). However, of AGPs likely play a role that from neutral of extensin. (Lamport et al., the and could (Tan et al., AGPs to as a We know that C AGPs as soluble AGPs that are then into the growing wall as putative pectic (Lamport, Lamport et al., We also know that the Yariv expansion growth and 1996) and that this with the role of AGPs as a in muro and expansion what we know about extensins as and AGPs as putative one extensins and AGPs as the and of cell and This (Kieliszewski and Lamport, et al., 1996) the of their biological they have a role that does not to be cross-linked or are they features of The first extensins, by formed a wall of Hyp-rich glycoproteins, notably and secreted to the cell in early a cleavage that the cell plate of later where of with their Hyp-rich cross-linking extensin a glycoprotein network or by the further addition of and pectic an extensin network with cross-links was a and led to walls that enabled and extensins are in plants as the maize THRGP cross-linked by extensin et al., 1996) and tobacco extensin the of and et al., Both are amphiphiles and therefore, with the potential for (Fig. of a role for extensins, they may of the cell plate while cross-linking extensins the of the primary cell wall. The of extensin in the first plants a role for extensins in with the earlier suggestion of as a protein (Smith et al., 1986). is that cross-linking extensins can now be to both of the first and notably in extensin of and a extensin in the et al., Both of the motif first in tryptic peptides isolated from tomato cell as extensin et al., 1986; Showalter et al., 1991). Idt motifs of a extensin are cross-linked in vitro by extensin peroxidase to the Tyr et al., in all some plants relatively Hyp of the a suggests that some notably have mechanical largely involving structural proteins (Kieliszewski and Lamport, 1987; Kieliszewski et al., 1990) but the primary role of extensin as a self-assembling with a role at Arabidopsis extensins are the with up to potential extensins, of which are very likely extensins et al., are potential extensins, and are extensin and extensins et al., 2010). extensins to or with differential extensin from their it is that extensins and proteins are to the et al., and et al., this includes et al., which also a a of extensin and protein et al., Furthermore, the identification of gene et al., is with the suggestion that of specific structural proteins a for of cell wall and 1989). The roles of extensins in cell wall cell and disease and in and et al., the of for each extensin specific extensins (Merkouropoulos et al., is a between extensin and walls that as et al., 1996), et al., and that cells to of they an of extensin et al., 1997). of extensins with extension growth is and 1967; and Chrispeels, and Shirsat, et al., 2009). The the of functional structures for one role are then later to different While it is the of this review to all extensin and (Solanum is a of the extensin motif to as a between two et al., are highly basic, and of the motif Averyhart-Fullard et al., 1988; Tierney et al., 1988; et al., and are cross-linked into the wall network et al., 1992; et al., direct evidence is Nevertheless, are with particular cell by and cells et al., work and that and are for cell function and 2010). Significantly, of structure and that the motifs of and a of This suggests that an protein contributes to the of cell The hypothesis contiguous Hyp residues as of and noncontiguous Hyp residues as of readily data from extensins and AGPs (Shpak et al., et al., 2002; et al., as AGP and extensin can be to glycosylation when in noncontiguous Hyp residues are the of the noncontiguous residues in AGPs can be in that the noncontiguous Hyp residues can be by as many as three or four residues and be of et al., 2002). Some and maize THRGP also noncontiguous Hyp residues or are example, HRGP is a PRP that only on contiguous Hyp while the major peptide motif also noncontiguous residues that are yet are other some are of the that also noncontiguous Hyp, with only four amino the Hyp residues that also et al., maize THRGP possesses repeats that some dicot extensins, like tomato P1, also noncontiguous residues that are work that of noncontiguous Hyp is by amino acid and repeats were and whereas and repeats were of and (Tan et al., 2003). Thus, the and amino acid and sequences of extensins and in and low in Ala and with the AGPs in Ala and low in Tyr and also that sequence in addition to the of Hyp residues a single Hyp is or work the Hyp contiguity The has conserved et al., 1972; et al., Thus, the of the Ser-Hyp4 a yet the is and does have potential and a unique the Ser residues are and all Hyp residues are with three to residues The structures are In to the and and of the of the Ser-Hyp4 are all the the which is et al., Thus, and a unique for including lectins or A rapid in muro response to to the insolubilization of extensins et al., at the level by of specific extensins containing Idt motifs et al., 1991; et al., 1992; et al., This a highly cross-linked network by extensin peroxidase and oxygen by fungal (Brady and Fry, 1997). Such a between disease resistance and extensin to be by molecular This review by to the initial that molecular oxygen is the direct source of the Hyp hydroxyl and may have a because the of is than that of it now to the of a direct of oxygen (Lamport, 1963). Such oxygen by specific is in animals et al., The corresponding oxygen of plants but plants also plant et al., therefore, we their likely role as an oxygen in the of and that to by et al., and